This is a proposal for competitive renewal of the currently funded R01 grant entitled Systematic psychophysical investigation of visual learning. Visual perceptual learning (VPL) is regarded as a promising tool that can help clarify important aspects of visual and brain plasticity. Our long-term goal is to delineate general rules that govern VPL, and to gain a better understanding of the mechanisms of visual and brain plasticity. During the currently supported period, we have attained great success in clarifying the characteristics of perceptual and behavioral changes under VPL, often by examining how performance at or around a trained feature value is changed as a result of visual training. However, new serious controversies about the mechanisms of VPL have emerged. One controversy concerns the locus of VPL, or the visual and brain information processing stage that is changed in association with VPL. The other concerns how VPL with location specificity or transfer is developed during training. We will conduct systematic psychophysical research with the aim of building two separate but related models, each of which concerns where or how VPL is developed to resolve each of these controversies, so that these two models will be integrated to a unified model of VPL. In Aim 1, we will attempt to resolve the controversy about the locus of VPL by developing a two-stage model that can account for both the findings that support low-level changes and those that support higher-level stages, without denying either evidence for the low-level or higher-level models. By conducting a variety of experiments with different types of tasks, stimuli and procedures, we will psychophysically test the validity of this two-stage model as well as existing models. In Aim 2, we will attempt to resolve the second controversy. The reduction of the degree of location specificity by double training has exposed serious limitations in how existing models explain location specificity. Thus far, no theoretical model has been proposed that can clearly explain how location specificity occurs by single training, as well as how location specificity is weakened/eliminated by double training. Based on our recent finding of elimination of VPL followed by reactivation and our influential reinforcement model, we will build the reinforcement reactivation model that assumes that the elimination of location specificity after double training is due to a failure to reconsolidate location specific learning tat was reactivated by reinforcement signals. In a series of experiments, we will test the validity of this reinforcement reactivation model and other existing models. In both aims, results of experiments will also be used to further refine the proposed models to most suitably account for where and how VPL is developed. Examinations of these aims will help to achieve our long-term goal of clarifying general functional rules that govern VPL.